Analysis of on-off patterns in VoIP and t heir effect on voice traffic aggregation Wenyu Jiang, Henning Schulzrinne Department of Computer Science Columbia U niversity Computer Communications and Networks, 2000. Proceedings. Ninth International Conference on Estimation of Token Bucket Parameters of VoIP Traffic R. Bruno, R.G.Garroppo and S.Giordano Department of Information Engineering Uni versity of Pisa High Performance Switching and routing, 2000. ATM 2000 Proceed ings of the IEEE Conference on , 2000
28
Embed
Analysis of on-off patterns in VoIP and their effect on voice traffic aggregation Wenyu Jiang, Henning Schulzrinne Department of Computer Science Columbia.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Analysis of on-off patterns in VoIP and their effect on voice traffic aggregation
Wenyu Jiang, Henning SchulzrinneDepartment of Computer Science Columbia University
Computer Communications and Networks, 2000. Proceedings. Ninth International Conference on
Estimation of Token Bucket Parameters of VoIP Traffic
R. Bruno, R.G.Garroppo and S.GiordanoDepartment of Information Engineering University of Pisa
High Performance Switching and routing, 2000. ATM 2000 Proceedings of the IEEE Conference on , 2000
Outline
• Introduction
• Experiment Setup
• Comparisons with Traditional Silence Detectors
• Token bucket simulations and results
• Conclusions
Introduction
• Human speech consists of talk-spurts and silence gaps, also known as on-off patterns.
• Allows higher bandwidth utilization through multiplexing.
• Allows per-spurt play out delay adjustment.
• Enable echo suppression based on silence detector output.
Introduction (cont’d)
energy
Pre-spurt hangover time
Post-spurt hangover time
time
Max (-20db)
Min (-45db)
Experiment Setup
3COM Ethernet Phone
3COM Ethernet Phone
Mediatrix gateway
Comparisons with Traditional Silence Detectors
• Example spurt/gap distributions
• NeVoT SD spurt and gap CDF using different parameters
• Spurt/gap distribution after averaging over many conversations
Example spurt/gap distributions
Example spurt/gap distributions
Example spurt/gap distributions
NeVot SD spurt and gap CDF using different thresholds
NeVot SD spurt and gap CDF using different thresholds
NeVot SD spurt and gap CDF using different thresholds
Spurt/gap distribution after averaging over many conversations
Spurt/gap distribution after averaging over many conversations
Spurt/gap distribution after averaging over many conversations
Token bucket simulations and results
Effect of spurt/gap distribution on multiplexing performance, G.729B
Effect of spurt/gap distribution on multiplexing performance, G.729B
Effect of spurt/gap distribution on multiplexing performance, G.729B
Multiplexing performance for NeVoT SD with default parameters
Multiplexing performance for NeVoT SD with default parameters
Multiplexing performance for NeVoT SD with default parameters
Equivalent Queuing Model
Birth-and-death Markov Chain
Simulation analysis
classifier token-bucket conditioner
Multiplexer
Conclusions
• Spurt/gap distributions are not exactly exponential, particularly for gaps.
• The token bucket simulations result indicate that the exponential model generally gives a close estimate of the out-of-profile probability.